An approach to optimize the path of humanoid robots using a hybridized regression-adaptive particle swarm optimization-adaptive ant colony optimization method

Abstract

Purpose

With enhanced use of humanoids in demanding sectors of industrial automation and smart manufacturing, navigation and path planning of humanoid forms have become the centre of attraction for robotics practitioners. This paper aims to focus on the development and implementation of a hybrid intelligent methodology to generate an optimal path for humanoid robots using regression analysis, adaptive particle swarm optimization and adaptive ant colony optimization techniques.

Design/methodology/approach

Sensory information regarding obstacle distances are fed to the regression controller, and an interim turning angle is obtained as the initial output. Adaptive particle swarm optimization technique is used to tune the governing parameter of adaptive ant colony optimization technique. The final output is generated by using the initial output of regression controller and tuned parameter from adaptive particle swarm optimization as inputs to the adaptive ant colony optimization technique along with other regular inputs. The final turning angle calculated from the hybrid controller is subsequently used by the humanoids to negotiate with obstacles present in the environment.

Findings

As the current investigation deals with the navigational analysis of single as well as multiple humanoids, a Petri-Net model has been combined with the proposed hybrid controller to avoid inter-collision that may happen in navigation of multiple humanoids. The hybridized controller is tested in simulation and experimental platforms with comparison of navigational parameters. The results obtained from both the platforms are found to be in coherence with each other. Finally, an assessment of the current technique with other existing navigational model reveals a performance improvement.

Research limitations/implications

The proposed hybrid controller provides satisfactory results for navigational analysis of single as well as multiple humanoids. However, the developed hybrid scheme can also be attempted with use of other smart algorithms.

Practical implications

Humanoid navigation is the present talk of the town, as its use is widespread to multiple sectors such as industrial automation, medical assistance, manufacturing sectors and entertainment. It can also be used in space and defence applications.

Social implications

This approach towards path planning can be very much helpful for navigating multiple forms of humanoids to assist in daily life needs of older adults and can also be a friendly tool for children.

Originality/value

Humanoid navigation has always been tricky and challenging. In the current work, a novel hybrid methodology of navigational analysis has been proposed for single and multiple humanoid robots, which is rarely reported in the existing literature. The developed navigational plan is verified through testing in simulation and experimental platforms. The results obtained from both the platforms are assessed against each other in terms of selected navigational parameters with observation of minimal error limits and close agreement. Finally, the proposed hybrid scheme is also evaluated against other existing navigational models, and significant performance improvements have been observed.

Citation

Kumar, P.B., Parhi, D.R. and Sahu, C. (2019), "An approach to optimize the path of humanoid robots using a hybridized regression-adaptive particle swarm optimization-adaptive ant colony optimization method", Industrial Robot, Vol. 46 No. 1, pp. 104-117. https://doi.org/10.1108/IR-10-2018-0204